Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Rovio feature tracking configuration #186

Open
npal1008 opened this issue Jun 28, 2018 · 4 comments
Open

Rovio feature tracking configuration #186

npal1008 opened this issue Jun 28, 2018 · 4 comments

Comments

@npal1008
Copy link

npal1008 commented Jun 28, 2018

Hi all,

I am trying to setup a simple hardware in order to use rovio as odometry software in real time. My goal is to achieve a stable odometry and the best possible with my hardware.
Currently I am using a See3Cam_CU135 with fisheye lens (154°) and a BNO055 based IMU. Since the camera accepts external hardware trigger, I setup a microcontroller that reads the IMU measurement via I2C and triggers the camera in order to achieve better timesync.

Now I have to adjust the config parameters of ROVIO which is quite difficult. The camera intrinsics and the camera-IMU extrinsics are quite good, but all other parameters (like covariance, noise, imgupdate etc) are not perfect, because my odometry isn't good (I saw that ROVIO can do better).

I especially noticed that the feature tracking is not so good. It looses often features resulting in using more than 6000 features in few meters. I think that this can be disturbing for the odometry

Are there any guidelines or tutorials or calibration processes that can help to find the best parameters for my configuration?

Thank you in advance

Edit: currently my Imu publishes at 100 Hz and the camera at 20 Hz

@npal1008 npal1008 changed the title Rovio configuration guidelines Rovio feature tracking configuration Jul 2, 2018
@npal1008
Copy link
Author

I did some test. Is it possible that the poor feature tracking result is due to a rolling shutter effect?

@bloesch
Copy link
Contributor

bloesch commented Aug 6, 2018

Rolling shutter could disturb the tracking, but if you move slowly that should be negligible. If you have more than 6000 features for a few meters then something seems wrong. Are you sure about the extrinsics/intrisics? The standard noise parameters usually work for various hardware setups. One thing you could also do is increase the IMU related noise parameters (vel, acb, gyb, att).

@npal1008
Copy link
Author

npal1008 commented Aug 20, 2018

How about the timing? I am using a configuration very similar to the one described in this example:

grauonline.de/wordpress/?page_id=1951

I am using the hardware trigger function of the camera, but maybe I am doing something wrong, because if I remove the trigger the results are better.

Is there a way to check how good the timing is?

@bloesch
Copy link
Contributor

bloesch commented Aug 28, 2018

If you remove the trigger the results get better something is probably wrong with the triggering setup. There is no rovio tool to check this, but you could https://github.com/ethz-asl/kalibr to calibrate the time offset. Also, be sure not to have any spurious timestamps at the beginning.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants